Among renewable energies, without considering hydroelectric energy, wind energy is among those that have demonstrated the greatest financial feasibility. Currently, the use of this type of energy has become globalized, and several European countries even plan to base their energy matrix on wind farms. This has been possible thanks to the application of advances in different fields of science and engineering that have allowed the construction of wind turbines (WTs) with increasing power and lower cost. However, one of the key factors to guarantee the viability of the wind industry is to minimize operating costs, especially those related to maintenance during the useful life of wind farms. In relation to maintenance, the main strategy of the wind industry is predictive maintenance based on the constant monitoring of various types of signals obtained from the components of the WTs by means of sensors. As all dynamic equipment produces acoustic or ultrasound vibration, this type of signal is the most widely used for monitoring from the blades to the tower, and most of the existing references on fault detection and diagnosis use the vibration signal. However, there is a lack of publications on other types of signals, especially when it comes to field work. Due to what has been stated so far, this thesis is X ÍNDICE